UK banks seeking access to Anthropic’s cybersecurity AI model, Mythos, remain largely on the waiting list, with only a small number – largely the UK arms of US lenders – having gained access so far, even as big British banks report no clear timeline. Anthropic received US permission to release the model to “trusted” US organizations and is coordinating with the US government as it expands availability abroad. Starling Bank’s chief information officer Harriet Rees, the finance ministry’s AI “champion”, advocates for more domestic AI infrastructure, models, and talent. Rohit Dhawan, Lloyds Banking Group’s head of AI, has proposed designating major AI providers as “critical” third parties under financial regulators, similar to the UK’s push to supervise dominant US cloud firms. If deemed “critical”, UK regulators could demand resilience testing, clearer incident reporting, and proof of end-to-end controls, raising the cost of serving UK finance and favoring incumbents with mature compliance programs. Access to models like Mythos may increasingly depend on whether vendors meet bank-grade oversight requirements, potentially tilting bargaining power toward large US firms and hindering smaller, non-US AI companies from winning bank contracts.